Licensing Collapse!
Ah yes. It’s all happening. Not gonna say I saw this coming, but I saw this coming.
Back when I published my first versions of FreeCT I obsessed over the licensing to apply (FWIW, GPLv2). I wanted to publish software that others could use but wouldn’t be co-opted and resold without some sort of in-kind contribution. We also live in a perverse time where you also have to disclaim liability for everything because people can’t be trusted to not blow up their stuff and then blame you. Wow. What a time to be alive.
That was back in 2015 and 2016 when modern deep learning was still just a glimmer in Jensen’s eye. And, to be honest, that was the last time I really gave licensing too much deep thought. Now, I mostly don’t care, beyond not wanting to be sued for anything or bothered more than necessary.
Really, I haven’t been feeling too concerned about this new development, but
Licensing and copyright are fundamentally useless for most of us.
When I say most of us, I’m talking about us computer people. Not the business people who look at computers and online services as money-printing machines, but those of us that love building, making, seeing things run, and actually improving the situation. It makes perfect sense that we would want to also be able to benefit financially from our efforts, but in my experience, the moment you start pursuing that, you have increasingly less time for the part that actually makes it all worthwhile and fun.
In spirit, licensing and copyright are a good thing, but under our current system if you’re not ready to protect your claim, it’s effectively invalid. Protection under licensing and copyright requires legal enforcement which is generally too expensiveand too unpleasant for most of us and most open-source efforts. Tracking your software around the world also might require you to add telemetry or DRM and other complexity your code would otherwise be completely unburdened by, without the need to enforce licensing.
The recent kerfuffle over the Claude-driven rewrite of chardet (related github issue) and subsequent “re-licensing” has laid bare the truth of all of this in a new way.
Licensing terms like those in the GNU Public License (GPL) were supposed to protect us somewhat, but the ability to completely reimplement an API with little-to-no code overlap using AI and then strip the GPL off and re-release it under a different license makes me wonder: what part of software actually has value anymore? We can’t patent/protect API reimplementation (a good decision, in my opinion), but we also can’t protect against AI reimplementation at this point of our interfaces. Huh.
To me, this looks like a system that was broken (even before it got this bad) being just more obviously broken in a new and obvious way. The system didn’t really protect or support small projects. The very concept of an “IP moat” illustrates that you need to go into licensing and IP-related maneuvering with a legal fund warchest. Even if I thought I had a billion dollar idea sounds utterly miserable to spend my time on.
I feel scared of is some knee-jerk reaction in the short-term making the long term viability of open-source software even more broken and impossible. If you’re a small developer and/or researcher who isn’t making a real business play, then consider the idea that you can choose to not play the game. Perversely, this is also sort of legally ambiguous and untested. GPL is also a great way to signal your project’s values outwardly to the community, but does present some very real legal hurdles to adoption and with the AI rewrite issue, doesn’t seem to provide you any real protection from IP or market-share theft.
In my view, none of this is our community’s failing, it’s that we’re trying to coexist with/under an administrative system that fundamentally does not understand these platforms (computational, social) it is trying to regulate. Not only does it not understand, it is often actively hostile towards us and our ideas.
I’ll leave this for now on a quote from Bruce Schneier that feels ever more relevant with each passing day:
Trying to make bits uncopyable is like trying to make water not wet. The sooner people accept this, and build business models that take this into account, the sooner people will start making money again.
To me, these idea of trying to control software and IP usage through the legal system is as wild trying to make water not wet. It’s just not what these machines are or how they operate.
I think the ultimate truth here is that some people contribute to the commons, and there are many who will take from the commons without giving back. If you post your code publicly, there’s not very much you can do to constrain who will benefit or what others will “derive” from your work. Instead, maybe concentrate on how you’re getting to spend your time. Are you working on topics you care about? Is the project advancing in a direction that aligns with your goals? Are your users taking advantage of your licensing terms or respecting it? For now at least, you can decide who you work with and for.
2026-03-12 13:37:!!